Repository navigation
Wave 9 Multi-Mac Deployment — 48 Workers across 2-3 Macs #409
Description
Wave 9 Multi-Mac Deployment
Overview
Deploy Wave 9 training with S3 MultiObj configuration across multiple Mac machines. Each Mac runs a subset of 48 total workers.
Configuration File: .trinity/wave9_multi_mac.yaml
devices:
- id: 1
hostname: mac-1.local
workers_start: 1
workers_count: 16
- id: 2
hostname: mac-2.local
workers_start: 17
workers_count: 16
- id: 3
hostname: mac-3.local
workers_start: 33
workers_count: 16CLI Commands
# Initialize all devices (generates compose files)
tri farm local-wave9 multi-mac init
# Start all workers
tri farm local-wave9 multi-mac start
# Check status
tri farm local-wave9 multi-mac status
# Verify seeds and S3 MultiObj config
tri farm local-wave9 multi-mac verify
# Stop all workers
tri farm local-wave9 multi-mac stopPer-Mac Device Init (Alternative)
# Mac 1 (workers 1-16)
tri farm local-wave9 device-init --device-id 1 --workers-start 1 --workers-count 16
docker-compose -f deploy/docker/docker-compose.wave9-mac-1.yml up -d
# Mac 2 (workers 17-32)
tri farm local-wave9 device-init --device-id 2 --workers-start 17 --workers-count 16
docker-compose -f deploy/docker/docker-compose.wave9-mac-2.yml up -d
# Mac 3 (workers 33-48)
tri farm local-wave9 device-init --device-id 3 --workers-start 33 --workers-count 16
docker-compose -f deploy/docker/docker-compose.wave9-mac-3.yml up -dVerification
# Verify unique seeds (CRITICAL!)
docker ps --format "{{.Names}}" | grep wave9 | while read c; do
docker exec $c printenv HSLM_SEED
done | sort | uniq -d # Should be empty
# Verify S3 MultiObj profile
docker exec wave9-worker-1 printenv HSLM_PROFILE # s3multiobj
docker exec wave9-worker-1 printenv HSLM_NTP_WEIGHT # 0.5
docker exec wave9-worker-1 printenv HSLM_JEPA_ENABLED # trueDocumentation
Full guide: docs/wave9/multi_mac_deployment.md
Closes #418
Activity
gHashTag commentedon Mar 25, 2026
✅ Wave 9 Multi-Mac Deployment — COMPLETE
All CLI commands implemented and tested:
Infrastructure files:
Build: L0 ✅ L1 ✅ tri ✅
feat(tri-lang): Bit/Trit-Level Pattern Matching (#409)
docs(tri): TTC freeze documentation (#409)
feat(tri-lang): Bit/Trit-Level Pattern Matching (#409)
docs(tri): TTC freeze documentation (#409)